LMO2 (rhombotin-2) - a bridging LIM-only protein, higher in pro-B and MLL-rearranged B-ALL, anti-selective in KMT2A-rearranged lines, and a physical partner of the KDM5A module that NUP98::KDM5A retains
also called RBTN2 · RBTNL1 · RHOM2 · TTG2 · Rhombotin-2 · LIM domain only protein 2 · Cysteine-rich protein TTG-2 · T-cell translocation protein 2
LMO2 is a tiny protein with no DNA-binding ability of its own that glues other transcription factors together in blood stem cells, and switching it on by accident caused leukaemia in two children during a gene-therapy trial. It is somewhat higher in the earliest B-cell leukaemias including KMT2A-rearranged ones, but knocking it out does nothing to KMT2A-rearranged cells.
The same thing, for a scientist
LMO2 (P25791, 158 aa, 11p13) is a LIM-domain-only bridging protein with no DNA-binding domain, and a member of the **heptad** whose combinatorial binding defines the blood stem/progenitor regulatory programme [E1]. In B-ALL its expression is stage- and genotype-patterned: across 247 ALL samples measured by real-time RT-PCR, **pro-B cases - the immunophenotype of KMT2A::AFF1 infant disease - carry three-fold higher LMO2 than pre-B, common-CD10+ or mature subtypes, and cases with TEL-AML1 or MLL rearrangement carry two-fold higher LMO2 than BCR-ABL or hyperdiploid cases** [E2]. That is a real, quantified, human, B-lineage expression association with the index disease, and it is the strongest thing on this entry. It is also **not a dependency**: DepMap gives LMO2 a leukaemia-arm mean of -0.378 outside the index arm and only -0.400 inside its myeloid half - which is entirely two lines, OCI-AML2 at -1.135 and SHI-1 at -1.153 - and **+0.096 in the B-lineage half**, giving z -0.05 myeloid and **+1.04 B-lineage**; the genome-wide scan ranks LMO2 **17,205 of 17,728** on KMT2A-selectivity with a difference of +0.1205, i.e. anti-selective [E3]. And the prognostic direction is **favourable**: high LMO2 correlates with better overall survival in adult ALL and is an independent favourable factor in normal-karyotype B-ALL [E2]. Two further threads matter. LMO2 activation by retroviral enhancer insertion at its promoter caused clonal T-cell proliferation in two SCID-X1 gene-therapy patients [E4] - the definitive statement that LMO2 dosage alone is leukaemogenic in the right cell. And **LMO2's second LIM domain binds the last 69 amino acids of RBP2/KDM5A** [E5], a region UniProt maps to KDM5A 1623-1690, which **overlaps the C-terminal PHD3 finger (1607-1661) that is the retained module of the NUP98::KDM5A fusion** - a physical link between one of this beat's genes and a fusion this graph already carries, never tested in that context.
This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.
I have never read one of these pictures before
A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.
Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.
Anything in yellow sticks is a bound small molecule, which is usually the drug.
Expressed in / required in / maintains a state
- Expressed in. Highest at the most immature stages of lymphopoiesis; in B-ALL, three-fold higher in pro-B than in later subtypes, and two-fold higher in TEL-AML1 or MLL-rearranged cases than in BCR-ABL or hyperdiploid ones 2. Also erythroid and germinal-centre B cells and diffuse large B-cell lymphoma 2. Heptad member in blood stem/progenitor cells 1. This is the only gene on this beat with a quantified expression measurement in human KMT2A-rearranged B-ALL.
- Required in. Not in KMT2A-rearranged lines - anti-selective, positive gene effect in the B-lineage half 3. It IS a deep dependency in a minority of leukaemia lines (2.3% of all lines at or below -0.5; OCI-AML2 -1.135, SHI-1 -1.153 inside the index arm), and the comparator leukaemia arm's mean of -0.378 is the deepest of the six genes here - consistent with the T-ALL literature LMO2 is best known for, which this beat did not survey.
- Maintains a state. The immature, stem/progenitor-like state - which is why its expression tracks pro-B rather than pre-B 2, and why its aberrant activation in a thymocyte drives clonal proliferation 4. No experiment in a KMT2A-rearranged cell tests whether LMO2 maintains anything.
The three registers come apart cleanly here and in opposite directions, which is the point of separating them: expression is up in the index disease, dependency is down, and prognosis is favourable. Treating the expression datum as evidence of a cooperating role would invert the prognostic sign.
The KDM5A contact, and why it is worth someone's afternoon
In 1997, a yeast two-hybrid screen using the RBTN2/LMO2 LIM-domain region as bait pulled out retinoblastoma-binding protein 2 (RBP2, now KDM5A/JARID1A). The interaction was confirmed by in vitro binding and by co-immunoprecipitation; deletion analysis showed the second LIM domain of LMO2 is necessary and sufficient for binding to the last 69 amino acids of RBP2; and the combination of RBP2 and RBTN2 gave higher transcription in vitro than RBTN2 alone 5. UniProt curates this as KDM5A REGION 1623-1690, 'Interaction with LMO2', with PubMed:9129143 as the experimental evidence 6.
Why that is interesting in this graph. protein-kdm5a-domains places KDM5A's third PHD finger at 1607-1661, and fusion-nup98-kdm5a states that the fusion retains only that C-terminal PHD finger of KDM5A and discards the entire catalytic apparatus, and that mutating the PHD so it cannot bind H3K4me3 abolishes transformation. The LMO2-interaction region and the retained PHD3 overlap across residues 1623-1661. So the one module NUP98::KDM5A keeps is also the module LMO2 was shown to bind.
This is an observation, not a finding, and the entry grades it accordingly. The 1997 work is a two-hybrid plus co-IP in a T-cell context with no fusion and no leukaemia model; a 68-residue UniProt region and a 55-residue PHD finger overlapping is a coordinate coincidence until someone does a pull-down; and the fusion's retained fragment may fold or be occluded differently from full-length KDM5A. It is recorded because it is cheap to test, because nobody appears to have noticed it, and because NUP98::KDM5A is a HOXA-high fusion in paediatric AMKL where an LMO2 contact would be mechanistically live.
The prognostic sign is favourable, which the cooperating-factor premise does not predict
High LMO2 expression correlated with better overall survival in adult ALL (5-year survival 64.8% versus 25.8%, P = 0.001) and was an independent favourable prognostic factor in normal-karyotype B-ALL (80.3% versus 63.0%, P = 0.043) 2.
A factor that a leukaemia driver co-opts to maintain itself does not usually predict a better outcome. The authors' own reading is that LMO2 expression in B-ALL reflects differentiation stage - the more immature the blast, the higher the LMO2 - and that the prognostic signal rides on whatever else that stage carries. Read that way, the MLL-rearranged association in 2 is a restatement of the pro-B immunophenotype of MLL-rearranged B-ALL, not an independent statement about the fusion. That is the collinearity problem this beat was sent to watch for, appearing in a different guise from the HOXA-cluster version.
Identifier hazards
- Reading the protein name back fails.
gene_exact:LMO2returns entry name RBTN2_HUMAN with recommended protein name 'Rhombotin-2'. Neither string contains 'LMO2'. A reader applying rule 65 concludes the lookup failed and may discard a correct record - the DPF2 failure shape, and the graph's own symbol-collisions file already logs DPF2 for exactly this. - The length defence is gone. LMO2 is P25791, 158 aa; LMO1 is P25800, 156 aa - a 2 aa gap, tighter than every pair currently in
data/symbol-collisions.jsonexcept the zero-spread cases.verify-accessions.pycannot separate them at any usable tolerance. LMO1 is Rhombotin-1 / TTG-1 / RBTN1, also a T-ALL oncogene, also a LIM-only bridging protein, and the two are discussed in the same sentences. - The previous official symbol is RBTNL1 and the aliases RBTN2, RHOM2 and TTG2 carry essentially all the pre-2000 literature, including 5 - which publishes under rhombotin-2 and RBTN2 and is unreachable from a bare LMO2 search of its title.
LMO2[tiab]returns 673 records;LMO2[tiab] AND (MLL OR KMT2A)[tiab]returns 13, of which the recent ones are T-ALL classifier and infant-T-ALL oncogenomics papers rather than KMT2A-cofactor work.
| uniprot | P25791 |
|---|---|
| gene symbol | LMO2 |
| length AA | 158 |
| alphafold ID | AF-P25791-F1 |
| ensembl | ENSG00000135363 |
| hgnc | HGNC:6642 |
| cytoband | 11p13 |
| protein class | LIM-domain-only bridging protein; no DNA-binding domain |
| resolution note | gene_exact:LMO2 AND organism_id:9606 AND reviewed:true returned exactly ONE reviewed row, primary gene name LMO2, ENTRY NAME RBTN2_HUMAN, PROTEIN NAME 'Rhombotin-2', 158 aa, synonyms RBTN2 RBTNL1 RHOM2 TTG2 - fetched live from the UniProt REST API 2026-09-20 with gene_primary and length requested and read back. Note that neither the entry name nor the protein name contains 'LMO2', so rule 65's read-the-name-back check LOOKS like a failed lookup on a correct record. Ensembl id from HGNC's single primary ensembl_gene_id (HGNC:6642), per rule 78. |
| paralogue discrimination | LMO2 P25791 (158 aa) against LMO1 P25800 (156 aa) is a TWO-residue gap - tighter than every non-zero pair currently in data/symbol-collisions.json, and far inside rule 66's 16 aa limit. verify-accessions.py offers no protection. LMO1 is Rhombotin-1 / TTG-1 / RBTN1, also a LIM-only T-ALL oncogene, discussed in the same literature. Only gene_primary discriminates, and the accessions differ by nine (P25791 / P25800). |
| domains note | Read back from the UniProt flat file for P25791 on 2026-09-20 (FT DOMAIN 30..89 and 94..153). LIM2 is the domain shown to be necessary and sufficient for the RBP2/KDM5A contact [E5]; the boundaries quoted there are UniProt's, not the 1997 paper's, which used deletion constructs this entry did not read. |
| cell type | human pro-B, pre-B, common-CD10+ and mature B-ALL blasts (expression gradient) [E2]; blood stem/progenitor cells [E1]; human T-lymphoid progenitors (insertional activation) [E4]; erythroid cells and germinal-centre B cells [E2]. |
| cell state | the immature/stem-progenitor-like state - LMO2 is highest at the most immature stages of lymphopoiesis and falls with maturation [E2]. No measurement of LMO2 across a differentiation or quiescence transition in a KMT2A-rearranged cell. |
| subcellular localization | Nucleus (UniProt P25791, ECO:0000305 - an inference from the record's curators rather than a direct experimental localisation code, which is worth noting for a 158-residue protein with no NLS annotated in the record read this cycle). |
| subnuclear localization | NOT ESTABLISHED in any context by anything read this cycle. LMO2 is a component of the SCL/TAL1-E2A-LDB1-GATA1 complex whose structures this graph already references (PDB 2YPA, 2YPB), but a structure is not a subnuclear address. |
| named but unwritten count at dispatch | 5 |
| named but unwritten count measured | 4 |
| what this entry does not claim | That LMO2 has a demonstrated functional role in KMT2A-rearranged leukaemia. It has none. The entry records a quantified expression association that is confounded by immunophenotype, a favourable prognostic direction, an anti-selective dependency profile with an unexplained bimodality, and an untested physical link to the KDM5A module that NUP98::KDM5A retains. |
domains
[
{
"name": "LIM zinc-binding 1",
"start": 30,
"end": 89
},
{
"name": "LIM zinc-binding 2",
"start": 94,
"end": 153
}
]shown for fusion
{
"in_kmt2a_rearranged_context": "EXPRESSION ONLY. Two-fold higher LMO2 in MLL-rearranged than in BCR-ABL or hyperdiploid B-ALL [E2], with the partner gene unresolved and the association confounded by pro-B immunophenotype. No functional experiment in any KMT2A-rearranged cell.",
"wild_type_only": "[E1] normal haematopoiesis; [E4] SCID-X1 gene therapy, insertional activation in T cells; [E5] yeast two-hybrid and co-IP with no fusion.",
"adjacent_fusion_lead": "[E6] - the curated LMO2-interaction region of KDM5A overlaps the module retained in NUP98::KDM5A. Untested."
}expressed required maintains
{
"expressed_in": "Pro-B B-ALL at three-fold the level of later subtypes; MLL-rearranged and TEL-AML1 cases at two-fold that of BCR-ABL and hyperdiploid [E2]. The best-quantified expression association with KMT2A-rearranged disease on this beat, and confounded by immunophenotype.",
"required_in": "NOT in KMT2A-rearranged lines - anti-selective, positive effect in the B-lineage half [E3]. Deeply required in a minority of leukaemia lines including two of seven KMT2A-rearranged myeloid ones (OCI-AML2, SHI-1), for reasons nobody has explained.",
"maintains_a_state": "Untested in this disease. The immature-progenitor association [E2] and the T-cell insertional-activation result [E4] both point at a stem/progenitor state, and neither is a KMT2A experiment."
}essentiality
{
"source": "DepMap CRISPR (Chronos) via Open Targets GraphQL, retrieved live 2026-09-20",
"n_all_lines": 1255,
"chronos_mean_all_lines": -0.0089,
"frac_lines_at_or_below_minus_0_5": 0.0231,
"mean_other_leukaemia": -0.3783,
"mean_kmt2a_r_myeloid": -0.4,
"z_kmt2a_r_myeloid": -0.05,
"kmt2a_r_myeloid_per_line": {
"OCI-AML2": -1.135,
"SHI-1": -1.153,
"MOLM-14": -0.306,
"THP-1": -0.105,
"MV4-11": -0.054,
"MOLM-13": -0.037,
"NOMO-1": -0.01
},
"mean_kmt2a_r_b_lineage": 0.0962,
"z_kmt2a_r_b_lineage": 1.04,
"kmt2a_r_b_lineage_per_line": {
"HB11;19": 0.323,
"SEMK2": 0.099,
"SEM": 0.045,
"RS4;11": -0.082
},
"positive_control_men1_z_myeloid": -2.41,
"genome_scan_rank": 17205,
"genome_scan_of": 17728,
"genome_scan_difference": 0.1205,
"interpretation": "anti-selective overall, and BIMODAL within the myeloid index arm - two of seven lines deeply dependent, five not. Per-line values are given because the mean is not a description of this distribution."
}depmap kmt2a selectivity
{
"chronos_mean_kmt2a_rearranged": -0.2196,
"chronos_mean_other_leukaemia": -0.3401,
"difference": 0.1205,
"pan_essentiality_all_lines": -0.0068,
"selectivity_rank": 17205,
"rank_denominator": 17728,
"provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
"how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}- LMO2 is 158 residues of two zinc-binding LIM domains with no enzymatic activity and no DNA binding - it is a pure protein-protein interface, which is the hardest class of target, and the interfaces it makes are with SCL/TAL1 and LDB1, both essential in normal haematopoiesis.
- Raising LMO2 is frankly leukaemogenic in humans: insertional enhancer activation at the LMO2 promoter produced clonal T-cell proliferation in two of ten SCID-X1 gene-therapy patients [E4]. Any vector-based therapy in this disease, including a CAR-T or a gene-edited graft, is working next to a locus with a documented human oncogenic insertion event.
- In B-ALL high LMO2 is a FAVOURABLE prognostic factor [E2], so lowering it has no supporting clinical rationale and a plausible adverse one.
- The anti-selective dependency profile means there is no KMT2A-rearranged potency window - and the two lines that ARE dependent are myeloid, so even a subset strategy would not reach the index B-lineage disease.
- Why are OCI-AML2 and SHI-1 deeply LMO2-dependent when the other five KMT2A-rearranged myeloid lines are not? Two of seven at effect sizes below -1.1 is not noise, and whatever distinguishes them is a co-dependency question answerable from public DepMap data.
- Does LMO2 bind the NUP98::KDM5A fusion protein? The curated LMO2-interaction region of KDM5A (1623-1690) overlaps the PHD3 module the fusion retains (1607-1661). One co-immunoprecipitation would settle it, and if positive it gives NUP98::KDM5A a bridging partner that connects it to the SCL/TAL1-LDB1 machinery.
- Is the MLL-rearranged LMO2 elevation anything other than the pro-B immunophenotype? The paper that reports it also reports the maturation gradient that would explain it. Comparing LMO2 within pro-B cases split by genotype - data that exist in every large B-ALL transcriptome cohort - would separate them.
- Why does high LMO2 predict BETTER survival in B-ALL when LMO2 activation is frankly leukaemogenic in T cells? The sign flip between lineages is unexplained and is the kind of thing that usually indicates the marker is reporting cell state rather than driving it.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
LMO2 is one of the seven transcription factors in the heptad - SCL/TAL1, LYL1, LMO2, GATA2, RUNX1, ERG and FLI-1 - identified by genome-wide binding analysis of ten major regulators as combinatorially interacting in blood stem/progenitor cells.
Combinatorial transcriptional control in blood stem/progenitor cells: genome-wide analysis of ten major transcriptional regulators · Wilson NK, Foster SD, Wang X, Knezevic K, Schütte J, Kaimakis P, Chilarska PM, Kinston S, et al. · Cell Stem Cell 7:532-544 2010What this evidence cannot show
Abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body not retrieved. NORMAL murine haematopoiesis, no fusion, no leukaemia. LMO2 has no DNA-binding domain, so its presence in a genome-wide binding set is necessarily indirect - it is there as a bridging partner of SCL/TAL1 and LDB1, and the abstract does not separate direct from bridged occupancy. Cited for heptad membership only.
Evidence that argues against this (1)
- All three heptad members on this beat - LMO2, ERG and GATA2 - are non-selective or anti-selective in the KMT2A-rearranged DepMap arm, with LMO2 at rank 17,205 of 17,728. Whatever the heptad is doing in normal stem cells, it is not producing a KMT2A-selective dependency. analysis run by this project 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In 247 acute lymphoblastic leukaemia patient samples with cytogenetic data measured by real-time RT-PCR, B-ALL cases expressed variable LMO2 depending on immunophenotype and cytogenetics: pro-B, the most immature subtype, displayed three-fold higher LMO2 than pre-B, common-CD10+ or mature subtypes, and cases with TEL-AML1 or MLL rearrangements exhibited two-fold higher LMO2 than cases with BCR-ABL rearrangement or hyperdiploid karyotype; clinically, high LMO2 correlated with better overall survival in adults (5-year 64.8% versus 25.8%, P = 0.001) and was a favourable independent prognostic factor in normal-karyotype B-ALL (80.3% versus 63.0%, P = 0.043).
LMO2 expression reflects the different stages of blast maturation and genetic features in B-cell acute lymphoblastic leukemia and predicts clinical outcome · Malumbres R, Fresquet V, Roman-Gomez J, Bobadilla M, Robles EF, Altobelli GG, Calasanz MJ, Smeland EB, et al. · Haematologica 96:980-986 2011What this evidence cannot show
Structured abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body (PMC3128216) not retrieved, so the number of MLL-rearranged cases within the 247, their partner genes and their ages are NOT known to this entry and are NOT asserted. HUMAN, B-LINEAGE, mixed adult and paediatric ALL; the 'MLL rearrangement' category is not resolved to a partner, so KMT2A::AFF1 specifically is not identified. THE CENTRAL CONFOUND IS STATED BY THE PAPER ITSELF: LMO2 expression tracks differentiation stage, and MLL-rearranged B-ALL is pro-B, so the two-fold genotype association may be entirely a restatement of the immunophenotype association. The survival analysis is in adults and in normal-karyotype cases, not in the MLL-rearranged group.
Evidence that argues against this (1)
- The prognostic direction is FAVOURABLE, which a cooperating-oncogene reading does not predict, and the paper attributes the expression pattern to maturation stage rather than to genotype. Both facts argue that the MLL-rearranged association is collinear with immunophenotype - the same class of confound as HOXA-cluster co-expression, which is the thing this beat exists to separate. Broad Institute DepMap; Open Targets Platform; query run by this project 2026-09-20 2026
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Fetched live on 2026-09-20, DepMap Chronos gene effect for LMO2 has mean -0.0089 across 1,255 cell lines with 2.31% of lines at or below -0.5; the 44 non-index leukaemia lines average -0.378, the KMT2A-rearranged myeloid half averages -0.400 (z -0.05) driven entirely by OCI-AML2 at -1.135 and SHI-1 at -1.153 with the other five between -0.010 and -0.306, and the KMT2A-rearranged B-lineage half averages +0.096 (z +1.04); the genome-wide scan ranks LMO2 17,205 of 17,728 on KMT2A-selectivity with a difference of +0.1205. MEN1 through the identical query shape returned z -2.41 / -2.43.
DepMap CRISPR (Chronos) gene effect for LMO2 (ENSG00000135363) and MEN1 (ENSG00000133895), retrieved via the Open Targets Platform GraphQL API, with the lineage split of analyses/lineage-confound-myeloid-tf-dependencies re-run for this gene · Broad Institute DepMap; Open Targets Platform; query run by this project 2026-09-20 · Database query 2026What this evidence cannot show
A bounded negative with a date, a route and a positive control through the identical query shape (rule 67). Seven myeloid and four B-lineage KMT2A-rearranged lines. THE MYELOID MEAN IS NOT A MEAN IN ANY USEFUL SENSE: two of seven lines are deeply dependent and five are not, so the arm is bimodal and reporting -0.400 as its central tendency would mislead - the per-line values are therefore given in the claim. The comparator leukaemia arm's -0.378 is itself likely driven by T-ALL lines, which this beat did not enumerate and which are where the LMO2 literature lives. Immortalised adult-derived lines, none infant. Ensembl id from HGNC primary (HGNC:6642); approvedSymbol read back as LMO2 so the LMO1 collision could not enter.
Evidence that argues against this (1)
- Two of seven KMT2A-rearranged myeloid lines ARE deeply LMO2-dependent, at effect sizes larger than MEN1's in some lines. A pooled z of -0.05 conceals that completely. Whatever distinguishes OCI-AML2 and SHI-1 from MOLM-13, MOLM-14, THP-1, NOMO-1 and MV4-11 is the interesting question, and the pooled statistic actively hides it - this is the floor-compression and pooling problem the standing brief warns about, appearing as bimodality rather than as compression. analysis run by this project 2026
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Almost three years after retrovirus-mediated gamma-c gene transfer into autologous CD34 bone marrow cells for SCID-X1, uncontrolled exponential clonal proliferation of mature T cells occurred in the two youngest of ten patients; both patients' clones showed retrovirus vector integration in proximity to the LMO2 proto-oncogene promoter, leading to aberrant transcription and expression of LMO2, most likely driven by retrovirus enhancer activity on the LMO2 promoter.
LMO2-associated clonal T cell proliferation in two patients after gene therapy for SCID-X1 · Hacein-Bey-Abina S, Von Kalle C, Schmidt M, McCormack MP, Wulffraat N, Leboulch P, Lim A, Osborne CS, et al. · Science 302:415-419 2003What this evidence cannot show
Abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body not retrieved. HUMAN, and it is the strongest causal statement anywhere that LMO2 dosage alone is leukaemogenic - but the cell is a T-lymphoid progenitor and the context is insertional enhancer activation in a gene-therapy setting, with no KMT2A fusion and no myeloid or B-lineage involvement. n = 2 events in 10 patients. Cited as the toxicity and dosage reference point, not as leukaemia-mechanism evidence for this disease.
Evidence that argues against this (1)
- LMO2 overexpression driving T-cell clonal proliferation in a thymocyte says nothing about whether LMO2 is required in a B-lineage or myeloid KMT2A-rearranged blast, and in B-ALL the direction of the clinical association is the opposite - high LMO2 predicts better survival. Cell context determines the sign, and this beat found no experiment establishing it in the index disease. Malumbres R et al. 2011
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
A yeast two-hybrid screen of a human cDNA library using the RBTN2/LMO2 LIM domain region as bait identified retinoblastoma-binding protein 2 (RBP2, now KDM5A/JARID1A) as a partner; the interaction was confirmed by in vitro binding assays and by co-immunoprecipitation of the two proteins; deletion analysis showed the second LIM domain of RBTN2 is necessary and sufficient for binding to the last 69 amino acids of RBP2; and the combination of RBP2 and RBTN2 gave higher transcription in vitro than RBTN2 alone.
T-cell oncogene rhombotin-2 interacts with retinoblastoma-binding protein 2 · Mao S, Neale GA, Goorha RM · Oncogene 14:1531-1539 1997What this evidence cannot show
Abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body not retrieved. A 1997 yeast two-hybrid plus in vitro binding plus co-IP, in a T-cell framing, with no leukaemia model, no fusion and no endogenous-complex demonstration. GRADED MODERATE for that reason. THE PAPER PUBLISHES UNDER 'rhombotin-2', 'RBTN2' and 'RBP2' AND IS UNREACHABLE FROM A TITLE SEARCH FOR LMO2 OR KDM5A - it was found only by reading the UniProt P29375 flat file, which is the route this entry recommends for old interaction literature. No claim about NUP98::KDM5A is made from it; the coordinate overlap noted in the sections is an observation about two curated regions, not a result.
Evidence that argues against this (1)
- A 1997 two-hybrid interaction that has, so far as this beat's searching found, never been followed up in twenty-eight years is weak ground. It survives in UniProt as a curated region with a single experimental evidence code, and a single-source curated interaction is exactly the kind of record this project has learned to treat as a lead rather than a fact. UniProt Consortium; record fetched by this project 2026-09-20 2026
- E6weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database
UniProt P29375 (KDM5A_HUMAN, 1,690 aa) annotates REGION 1623-1690 as 'Interaction with LMO2' with experimental evidence code ECO:0000269|PubMed:9129143, and this graph's protein-kdm5a-domains entry places KDM5A's third PHD finger at 1607-1661 while fusion-nup98-kdm5a states that the fusion retains only that C-terminal PHD finger - so the curated LMO2-interaction region and the retained module overlap across residues 1623-1661.
UniProtKB P29375 (KDM5A_HUMAN): FT REGION 1623..1690 /note='Interaction with LMO2' /evidence='ECO:0000269|PubMed:9129143'; SQ 1690 AA · UniProt Consortium; flat file fetched by this project 2026-09-20 · Database record 2026What this evidence cannot show
GRADED WEAK AND IT IS AN OBSERVATION, NOT A FINDING. The UniProt region was read back from the live flat file this cycle. The PHD3 coordinates (1607-1661) and the fusion's retained-module statement were read from THIS GRAPH's own entries - protein-kdm5a-domains and fusion-nup98-kdm5a - and were NOT independently re-derived from UniProt or from the fusion's founding paper this cycle, which is the weakest link in the chain and is stated so a reader does not treat the overlap as verified twice. Two curated regions overlapping is a coordinate coincidence until a pull-down is done: the fusion's retained fragment may fold or be occluded differently from full-length KDM5A, and nothing tests LMO2 binding to NUP98::KDM5A. No experiment anywhere connects LMO2 to any NUP98 fusion.
Evidence that argues against this (1)
- The retained KDM5A module in NUP98::KDM5A is required for transformation specifically as an H3K4me3 READER - PHD mutations that abolish H3K4me3 binding abolish transformation - so the function attributed to that region is nucleosome recognition, not protein bridging. An LMO2 contact overlapping it would have to be compatible with, or competitive with, H3K4me3 binding, and nothing addresses which. Mao S et al. 1997
NHLH1 is a member of the SCL/TAL-1 bHLH family whose leukaemogenic members work through the TAL1-LMO2 complex; NHLH1 itself is NOT implicated in T-ALL
- expressed in →Pro-B cell — the stage at which KMT2A::AFF1 leukaemia is arrested, and the phenotype the diagnosis is made on
three-fold higher LMO2 in pro-B B-ALL than in pre-B, common-CD10+ or mature subtypes
- expressed in →KMT2A-rearranged acute lymphoblastic leukaemia
two-fold higher LMO2 in MLL-rearranged and TEL-AML1 B-ALL than in BCR-ABL or hyperdiploid; confounded by pro-B immunophenotype
both members of the heptad of combinatorially acting blood stem/progenitor transcription factors
both members of the heptad of combinatorially acting blood stem/progenitor transcription factors
- related to →NUP98::KDM5A (NUP98-JARID1A), t(11;12)(p15;p13)
UNTESTED LEAD: the curated LMO2-interaction region of KDM5A (1623-1690) overlaps the PHD3 module (1607-1661) that this fusion retains
- expressed in →ETV6::RUNX1 (TEL-AML1), t(12;21)(p13;q22)
TEL-AML1 cases carry two-fold higher LMO2 than BCR-ABL or hyperdiploid cases, the same elevation seen in MLL-rearranged cases
LMO2 rank 17,205 of 17,728, difference +0.1205; the myeloid index arm is bimodal, two of seven lines below -1.1
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
ANTI-SELECTIVE: positive gene effect in the B-lineage half of the KMT2A-rearranged arm